To investigate empirical assumptions about substituted judgment, three treatment decision scenarios were presented to 43 competent elderly persons and 115 persons related to them. Related subjects who were explicitly asked to make a substituted judgment came significantly closer to the elderly person's preferences than those who were asked to make their best recommendation (p = .005–.067, depending on index used). These results support the use of a substituted judgment approach to proxy decision makers; other results suggest some limitations of durable powers of attorney.